[gmx-users] dummy atoms & timesteps

[Anton Feenstra]

J Simms wrote:
> 
> Hi all,
> Qu1I am just reading Antons '99 paper re large time-scale. Are the
> dummy atoms produced by pdb2gmx implimented the same as in
> the paper, or are they different?

Yes - exact same thing. I implemented it myself in pdb2gmx
explicitly for this paper. There is a more recent addition
to dummify aromatic amino-acid sidechains also, but the
article wich contains the corresponding description did not get
past the referees:

K. Anton Feenstra, Herman J. C. Berendsen and Alan E. Mark Hierarchical
      Motion in Small Proteins: Implications for Protein Folding . 
http://www.chem.vu.nl/staf/feenstra/articles/preprint2001.pdf


> Qu2. Can I just let pdb2gmx get on with producing dummy atoms,
> or do I have to look & do anything ;-)

Yes - the implementation in pdb2gmx for -dummy hydrogens at least
is designed to be generic, although I haven't tested with explicit 
hdyrogen forcefields.


> Qu3. When doing MD with dummy atoms, heavy hydrogens and a
> large timestep. What do I look for when comparing it with a smaller
> (ie 2fs) timestep & checking its right.

That basically depends on whatever properties you are interested
in. You could as a basic check look at e.g. rmsd values an so on.
I've used calculated NMR relaxation parameters as a comparison 
(these are not trivial to calculate, and not implemented in one of
the Gromacs analysis tools) in one of my recent papers on trajectories
run at 2 (normal) and 4 and 7 fs timestep (w/ heavy&dummy):

K. Anton Feenstra, C. Peter*, Ruud M. Scheek, Alan E. Mark and Wilfred F.
      van Gunsteren* A comparison of methods for calculating NMR
      cross-relaxation rates (NOESY and ROESY intensities) in small peptides.
      (2002) J. Biomol. NMR 23, 181-194. 
      * Laboratory of Physical Chemistry, Swiss Federal Institute of Technology
      Zuerich (ETHZ) 
http://www.chem.vu.nl/staf/feenstra/articles/jbnmr-23-3-181-194-(2002).pdf


> Qu4. Are there other paper re increasing the timestep of MD?

See references in my paper. I haven't checked the more recent literature.

I hadn't found any whith something similar to my dummy constructions.
Most modified the hydrogen masses, but without compensating the total
system mass. I (i.e. the pdb2gmx implementation) shift the increased
hydrogen mass from the bonded heavy atom (i.e. the heavy atom gets lighter).
Increasing total system mass scales your overal time-scale! 

One particularly bad one scaled the mass of *all* atoms in the system and 
keeping *velocities* the same. That is equivalent to taking a larger 
timestep (scales as sqrt(T_mass) ) and increasing temperature (which 
derives from 1/2 m*v^2).


-- 
Groetjes,

Anton
 ________ ___________________________________________________________
|        | Anton Feenstra                                            |
| .      | Dept. of Pharmacochemistry - Vrije Universiteit Amsterdam |
| |----  | De Boelelaan 1083 - 1081 HV Amsterdam - The Netherlands   |
| |----  | Tel: +31 20 44 47608 - Fax: +31 20 44 47610               |
| ' __   | Feenstra at chem.vu.nl - http://www.chem.vu.nl/afdelingen/FAR|
|  /  \  |-----------------------------------------------------------|
| (    ) | Dept. of Biophysical Chemistry - University of Groningen  |
|  \__/  | Nijenborgh 4 - 9747 AG Groningen - The Netherlands        |
|   __   | Tel +31 50 363 4327 - Fax +31 50 363 4800                 |
|  /  \  | K.A.Feenstra at chem.rug.nl - http://md.chem.rug.nl/~anton   |
| (    ) |-----------------------------------------------------------|
|  \__/  | "Three Little Fonzies" (Pulp Fiction)                     |
|________|___________________________________________________________|